Lanefan
Victoria Rules
Various people upthread have done a fine job of defining simulation in the same (or better!) terms that I'd use, which means I don't have to type it all out. 
In short (yeah, right!) and in summary, then, simulation to me represents a collection of elements mostly around relatability:
--- internal consistency within the setting*. That is to say, if something works in a particular way once then there's a reasonable expectation that, barring external influences, it'll work that way every time.
--- enough relatability to the real world that players can immediately grasp the basic stuff (gravity, time, weather, tides, human setting inhabitants, etc.) because they already have experience with those things in reality. Further, if any of those basic elements are to be inherently different in the setting (e.g. there's no gravity, or the weather never changes, or the humans are all seven feet tall) those things are clearly communicated up front so as to allow the players to adjust their imaginations accordingly.
--- using abstraction and rules to as best as possible+ mirror the setting's physics, rather than tweaking the setting's physics to suit the abstraction. And yes, this means that if the setting's physics are to be significantly different from our own, those differences need to be figured out before rules design even begins e.g. if faster-than-light travel is an element of the setting, how do the physics of it actually work.
--- the ability to explain how and why fantastic things work as they do, or at all, when compared with or added to our own reality; such that players can more easily relate to the fantastic. In other words, the physics of the fantastic get melded in with the physics we already know, thus making the simulation a more-cohesive whole. See above re faster-than-light travel.
* - or discrete parts thereof. For example, a D&D party journeying to a different plane of existence should perhaps expect basic things to work differently there than they do at home, while still reasonably expecting that new plane to have its own internal consistency once they become more familiar with the place.
+ - practicality rears its ugly head, of course, often requiring some trade-offs and compromises between accuracy of sim and ease of playability. This is the bane of anyone who has ever tried to design a deep-sim game.
In short (yeah, right!) and in summary, then, simulation to me represents a collection of elements mostly around relatability:
--- internal consistency within the setting*. That is to say, if something works in a particular way once then there's a reasonable expectation that, barring external influences, it'll work that way every time.
--- enough relatability to the real world that players can immediately grasp the basic stuff (gravity, time, weather, tides, human setting inhabitants, etc.) because they already have experience with those things in reality. Further, if any of those basic elements are to be inherently different in the setting (e.g. there's no gravity, or the weather never changes, or the humans are all seven feet tall) those things are clearly communicated up front so as to allow the players to adjust their imaginations accordingly.
--- using abstraction and rules to as best as possible+ mirror the setting's physics, rather than tweaking the setting's physics to suit the abstraction. And yes, this means that if the setting's physics are to be significantly different from our own, those differences need to be figured out before rules design even begins e.g. if faster-than-light travel is an element of the setting, how do the physics of it actually work.
--- the ability to explain how and why fantastic things work as they do, or at all, when compared with or added to our own reality; such that players can more easily relate to the fantastic. In other words, the physics of the fantastic get melded in with the physics we already know, thus making the simulation a more-cohesive whole. See above re faster-than-light travel.
* - or discrete parts thereof. For example, a D&D party journeying to a different plane of existence should perhaps expect basic things to work differently there than they do at home, while still reasonably expecting that new plane to have its own internal consistency once they become more familiar with the place.
+ - practicality rears its ugly head, of course, often requiring some trade-offs and compromises between accuracy of sim and ease of playability. This is the bane of anyone who has ever tried to design a deep-sim game.